Thank You very much Victor, I think I understand now how it can be done. If you could give me one last pointer, how to overwrite the verification callback function, that is called when executing "BIO_do_handshake", I'd be very grateful.
sorry for using misguiding vocanulary :) Michael On Thu, Sep 24, 2009 at 5:08 PM, Victor Duchovni < victor.ducho...@morganstanley.com> wrote: > On Thu, Sep 24, 2009 at 04:23:03PM +0200, Michael Prinzinger wrote: > > > > Are you saying that the accepting system expects X.509 client > credentials > > > from the connecting system, but that the payload (encrypted to the > > > receiving node's public key) also contains the same certificate, and > > > you want to check that the peer client matches the encrypted request? > > > > Yes. > > > > > If so, trust chain verification is completely irrelevant here. You > > > don't need to repeat the handshake, rather implement a verification > > > callback that accepts untrusted certificates (X.509 trust anchors seem > > > inapplicable in this context) and compares them (via the public key > > > fingerprint) to the designated certificate in the message payload. > > > > Yes, this would do what I want. > > I would have to call BIO_do_handshake with a customized verification call > > back, that decrypts the payload, reads the certificate and compares them > > to the one offered by the client? > > No, your verification callback just always succeeds, allowing invalid > trust chains, expired certificates, ... to be used. After the handshake, > when you receive and decrypt the payload, you compare the enclosed peer > certificate (public key fingerprint) with the peer certificate (public > key fingerprint) from the SSL session. > > > Since there are no central trust authorities withtin the context of this > > protocol, I can only directly verify a certificate, resp. some message > > signed with the corresponding private key. > > "Verify" is the wrong term here. It suggests X.509 PKI hierarchies and all > that jazz. You just want to validate the peer's client cert against the > message routing data in the payload. > > > > If you want something else, you must explain it in a lot more detail. > > > > > > Note, your problem is understanding the crypto protocol design, not > > > OpenSSL. You are using this list to get help with X.509 and > cryptography, > > > not really the OpenSSL API, so we are somewhat outside the list > charter, > > > but such questions are relatively common here... > > > > Thank you for helping anyways. > > If you know a good resource that explains X.509 infrastructure, I'd be > glad > > to read it. > > Actively avoid reading anything about X.509 infrastructure. Reading a > decent applied cryptography book + a decent book general about SSL/TLS > would be better (skipping the parts that deal with X.509 "infrastructure"). > > -- > Viktor. > ______________________________________________________________________ > OpenSSL Project http://www.openssl.org > User Support Mailing List openssl-users@openssl.org > Automated List Manager majord...@openssl.org >
